Elvin Li (李忞皓)

About Me

I'm a Stanford MS Computer Science student, and broadly enjoy systems and modeling in machine learning. I hope to create important things throughout my life. Philosophies I adhere by are to not separate the art from the artist, and to talk primarily through experience. I believe the best way to learn is to produce instead of consume.

Software

I've worked at Amazon Alexa and AWS Cryptography. In the former I built a distributed system that created service dependency graphs and predicted downstream traffic, fostering automatic scaling workflows for Alexa+ services. In the latter I implemented post-quantum cryptography to the AWS Encryption SDK via the ML-KEM algorithm.

Research

My current interests are in inference systems and ai safety. I've previously published on topics in representation learning, continual learning, and adversarial machine learning at AAAI, Computer Networks Journal, and various IEEE venues. My Erdős number is 4, which is possibly undeserved (for now).